MLA-C01 Question 415
Single answerYou are deploying a real-time machine learning inference application that processes thousands of requests per second with minimal latency. The model's inference process is computationally lightweight but requires high throughput and low latency. Which instance type would be most suitable for this use case?
- A
Compute optimized instances (e.g., C6g)
- B
Memory optimized instances (e.g., R6i)
- C
Inference optimized instances (e.g., Inf1)
- D
General purpose instances (e.g., M6g)
Show answer and explanation
Correct answer: C
Explanation
Inference optimized instances, such as Inf1, are specifically designed to handle machine learning inference workloads. They leverage AWS Inferentia chips, providing high throughput and low latency, which are critical for real-time inference applications. In this scenario, where the application needs to process thousands of requests per second with minimal latency, inference optimized instances are the ideal choice.
- A. Incorrect.
Compute optimized instances are designed for applications that require high compute power, such as high-performance computing or batch processing. They are not specifically tailored for inference workloads requiring low latency.
- B. Incorrect.
Memory optimized instances are suited for workloads requiring high memory, such as in-memory databases or large-scale data processing. They are not ideal for lightweight inference tasks.
- C. Correct.
Inference optimized instances are purpose-built for machine learning inference with high throughput and low latency. These instances, like Inf1, are the most suitable for real-time inference scenarios.
- D. Incorrect.
General purpose instances provide a balanced combination of compute, memory, and networking resources. While they can handle a variety of tasks, they are not optimized for inference-specific workloads.